Root production in a subtropical pasture is mediated by cultivar and defoliation severity

Background Grasslands occupy significant land area and account for a large proportion of the global soil carbon stock, yet the direct effects of grazing and genotypic composition on relationships between shoot and root production are poorly resolved. This lack of understanding hinders the development of models for predicting root production in managed grasslands, a critical variable for determining soil carbon stocks. Methods We quantified the effects of season-long defoliation treatments on both shoot and root production across four cultivars of a widely-planted pasture grass species (Paspalum notatum Fluegge) in a common garden setting in South Florida, USA. Results We found that infrequently applied (4 week) severe defoliation (to 5 cm) substantially enhanced shoot production for all cultivars, while severe defoliation reduced root production across cultivars, regardless of frequency. Overall, cultivars varied substantially in root production across the range of defoliation treatments in our study. However, there was no significant relationship between shoot and root production. Conclusions Our results find that aboveground and belowground productivity are only weakly coupled, suggesting caution against use of simple aboveground proxies to predict variations in root production in grasslands. More broadly, our results demonstrate that improved modeling and management of grasslands for belowground ecosystem services, including soil carbon sequestration/stocks, will need to account for intraspecific genetic variations and responses to defoliation management.


59
Grassland ecosystems occupy more than a fifth of earth's land area and account for a 60 large proportion of the global SOC stock (1,2). However, there is considerable uncertainty in 61 predictions of net ecosystem exchange, and hence carbon sequestration services from grasslands 62 (3,4). One significant source of uncertainty is that while large herbivore grazing is known to 63 mediate patterns of plant species composition, diversity, and aboveground primary productivity 64 (5-7), the effects of grazing on belowground processes and soil carbon is less clear (8-11). In 65 particular, there are limited field studies where the impact of grazing on root production in 66 grassland systems has been directly measured (e.g., via root ingrowth cores or minirhizotron 67 technology, but see Ziter and MacDougall (12) Balogianni et al. (11) and Cooley et al. (13)). 68 Since belowground production may be the largest component of total NPP for many grasslands 69 (14,15), determining how grazing affects root production will help to predict if and when 70 grassland ecosystems will behave as carbon sinks, and whether grazing is likely to promote or 71 inhibit carbon sequestration services. 72 Root carbon inputs may constitute a disproportionate amount of the total SOC stock 73 compared with shoot carbon (16-18), and are especially critical in grassland ecosystems where 74 aboveground tissue is susceptible to frequent removal by fire and grazing (19). Current 75 understanding of how grazing affects root production is ambiguous. For example, one temperate 76 mesocosm study showed that intense defoliation inhibited root production and accelerated the  Grazing effects on belowground production may not only vary based on plant species, but 90 also on the genotypic composition of a grazed stand, given the increasing evidence of the 91 importance of intraspecific variation in driving ecosystem structure and function (26,27). In 92 general, some literature suggests that reduced root allocation (and increased shoot allocation) 93 following grazing may represent an evolutionarily adaptive trait for grazing tolerance (28). For 94 instance, Carman (1985) (29) noted that short-leaved genotypes of Schizachyrium scoparium, 95 selected from a long-term grazed site, exhibited lower rates of root elongation post-grazing than 96 longer-leaved genotypes from a long-term grazing excluded site. Planted pasture grasses also 97 have been shown to exhibit genotypic variability in shoot and root production in response to 98 grazing (e.g. Dawson et al. (30) given the evidence for potentially significant genotypic and defoliation effects on belowground 113 carbon allocation, it is unclear whether aboveground proxies can ever reliably approximate root 114 production. Given the central importance of root system carbon inputs to maintaining SOC, 115 especially in grasslands, we need more data from experimental systems where genotypic 116 composition and grazing management have been manipulated, and the relationship between 117 above and belowground allocation have been quantified.

118
In this study, we tested the independent and combined effects of defoliation intensity and 119 frequency, and cultivar on root production of a widely utilized pasture grass species of the 120 southeastern United States, Paspalum notatum Flüegge (bahiagrass). For Bahiagrass, we can 121 broadly delineate cultivars on the basis of growth habit where historically older, widely 122 naturalized cultivars tend to have a more prostrate growth pattern, whereas recently selected 123 cultivars tend to have a more upright growth pattern, reflecting selection for improved forage 124 growth characteristics (38). Previous work, and considerable producer experience, suggests that 125 bahiagrass has a remarkable resilience to intense grazing, wherein forage growth and quality is 126 maximized with severe defoliation (close to ground level) so long as regrowth intervals are 127 adequate (39,40). However, the impact of defoliation severity on root production across 128 cultivars, and their associated growth habits, has not been directly studied, reflecting a general 129 lack of information on belowground growth responses in warm season subtropical pasture (13).

130
To redress this gap in knowledge we conducted an experiment in a common garden setting under 131 realistic conditions of limited soil fertility to isolate the effects of defoliation intensity, frequency 132 and cultivar on belowground production, and to evaluate the relationship between aboveground 133 and belowground growth. precipitation, and evapotranspiration, and all fell within normal ranges (Table A1). 170 We initiated defoliation treatments on June 13 th 2013 and concluded field sampling 16 171 weeks later on October 5 th 2013. Although we did not measure soil moisture, the soils were all 172 visibly waterlogged from July until the end of the experiment, as is typical in Florida Spodosol soils (43). We therefore assumed that plant growth was not limited by water availability during 174 the sampling period, or at the very least that water availability was essentially constant across 175 plots. Each plot (n = 32) was randomly assigned to either a frequent (2 week) or infrequent (4 176 week) defoliation treatment to simulate grazing stress. Each plot was divided in half and received 177 two defoliation intensities (a severe defoliation to 5 cm residual height, and a mild defoliation to 178 15 cm residual height) resulting in n = 64 experimental units ( Figure A1). Thus, our design was 179 effectively split-plot with two main-plot treatments (cultivar and defoliation frequency), while 180 our subplot factor was defoliation intensity. Overall, each cultivar X defoliation severity X 181 defoliation frequency treatment was replicated 4 times.

182
We harvested a 0.92-m 2 quadrat from each subplot during each defoliation treatment with  Aboveground production values are presented in gm -2 (dry biomass).

191
To quantify root primary production in response to the defoliation treatments, we observable pattern (Fig 1) that it is the combination of severe + infrequent (4 wk) defoliation that 281 leads to over-yielding. Overall, we did not estimate substantial variability in shoot production 282 among cultivars across all treatments, although the upright cultivars (UF-Riata and Tifton-9) had 283 slightly higher production than the decumbent cultivars Argentine and Pensacola (Fig. 3a).  Root production model 308 We observed an average root production of 224 gm -2 , where mild defoliation treatments 309 were the highest with 262 gm -2 averaged across 2 wk and 4 wk defoliation frequencies, 310 compared with severe defoliation with an average of 186 gm -2 (Fig 1). The fixed main effect 311 estimate for severe defoliation was negative (-0.33 +/-0.12, Fig 2b), with >97.5% of posterior 312 probability below 0, while the main effects of frequent defoliation and the interaction of frequent 313 X severe defoliation were highly uncertain, with 95% credible intervals spanning a similar range 314 above and below zero. Average root production across all treatment groups varied by cultivar 315 more substantially than shoot production (Fig 3b), with the decumbent cultivars Argentine and 316 Pensacola having greater root production than the upright cultivars UF-Riata and Tifton-9 (Fig   317   3b, Fig. 4). The greatest contrast was between Argentine and UF-Riata, which had a median 318 posterior difference of -0.36 on the log-link scale (Fig. 4), which represents a 30% lower root 319 production. Plots include a median (point), and 50% (thick line) and 95% (thin line) credible intervals.

323
Where the entire 95% credible interval falls above or below zero, we can interpret that as a 324 97.5+% Bayesian probability of the first cultivar having a higher root allocation than the second 325 cultivar.

327
Root allocation 328 The fixed main effect estimate for severe defoliation on root allocation proportion was -329 0.34 +/-0.09 (Fig 2c), a very similar median estimate to that for root production, although with a 330 smaller uncertainty (SE = 0.09 versus 0.12). This result represents a median estimate of 29% 331 reduced allocation proportion to roots overall among cultivars and across both frequencies of 332 defoliation with severe defoliation. Variation among cultivars was also similar to that observed 333 for root production (Fig 3c versus 3b), and thus we did not repeat the pairwise analysis since it 334 would convey redundant information.

336
Root production predictions 337 The univariate regression between shoot and root production revealed a very weak (R 2 = 338 0.09) relationship (Fig 5a). The full model that included treatment indicators and cultivar identity 339 (as in the analyses above), yielded a median R 2 of 0.45 (Fig 5b). After removing the varying predicted versus observed scatterplot for root production as predicted by defoliation treatment, 347 cultivar identity, and shoot production. For reference, the 1:1 line of "perfect fit" is plotted along 348 with an in-sample median Bayesian R 2 for both predictive models.

352
Severe defoliation resulted in substantially greater shoot production when applied 353 infrequently, but reduced root production among the bahiagrass cultivars. Averaged across all 354 defoliation treatments, root production was also more strongly variable among cultivars than was  Fig 1), 390 root production was not markedly lower than in our severe + infrequent treatment. Overall, it 391 appears that in our system, severity, not frequency, of grazing is the more important determinant 392 of grass root production. 393 We observed substantial overall variability in root production among the grass cultivars. in belowground production was unexplained, even in our best model, suggesting significant 420 spatial heterogeneity in root system productivity that should be further investigated. Root production is critical for maintaining and increasing soil carbon pools in grassland 456 ecosystems, yet findings on the immediate and long-term effects of grazing on root production 457 remain variable. We hypothesized that severe defoliation, if applied infrequently, might lead to 458 overyielding of shoots, but would have only small impacts on root production. Moreover, we 459 hypothesized that cultivars selected for an upright growth habit would show less root production 460 overall, and would be more sensitive to defoliation stress. Overall, we found that severe 461 defoliation per se, regardless of frequency, suppressed root production, even as infrequently 462 applied severe defoliation increased shoot production. Thus, it appears that manipulating timing 463 and intensity of grazing to optimize forage production might evoke a negative tradeoff with root 464 production. We did find support for the hypothesis that recently developed upright cultivars have 465 lower root production, and a lower root:shoot ratio, than widely naturalized decumbent cultivars.

466
The main limitation of our work is that realistic animal grazing management can differ from 467 experimentally imposed defoliation in two major ways: 1) grazing impacts will fall along a